Decision markets are increasingly common to help weed out the best decisions, but they are still dwarfed by decisions made via committees or other adhoc groups. It’s a choice that new research from UC Riverside suggests is increasingly untenable.
Despite this, the research finds that people appear to trust group-based decisions, even though they’re so susceptible to manipulation and poor decision making. However, the researchers find that people can learn to trust information markets the more often they use them, thus offering a ray of light.
“Our key finding was that transparency and trust are why people prefer groups even though markets outperform them,” the researchers say. “People are skeptical of algorithms. Markets will be used less until people get used to them.”
Trusting the crowd
Prediction markets work by participants placing bets on particular outcomes. Such markets have proven themselves to be better predictors of everything from elections to the Oscars, as they tap into the ‘wisdom of crowds’ and aggregate the diverse perspective of the participants.
By contrast, groups and committees usually have a much smaller pool of participants, often with homogenous knowledge. They’re also easier to manipulate, with conflicts among members common, and a greater likelihood that decisions will bend to the will of the most powerful members.
When deployed within companies, prediction markets often strive for involvement from across the organization, with each participant given the same amount of ‘money’ to place their bets on the various outcomes contained within the market.
The researchers believe that prediction markets tend to elicit our honest input precisely because we’re asked to put our money where our mouth is. What’s more, our ‘bets’ don’t require us to align ourselves with our organizational goals.
Limited use
Despite this seemingly robust logic, the use of prediction markets is relatively limited. The researchers conducted a number of experiments to figure out why this might be so. In one, for instance, participants were asked to select someone for a managerial position, with the choice made via either a face-to-face group or an online prediction market. The participants were given a range of roles and financial incentives, all of which were manipulated by the researchers.
The results revealed that groups can perform pretty well when the members have common incentives and interests, and can even outperform prediction markets. This isn’t the case when there are conflicts of interest among members however. Interestingly, these conflicts were often not spotted by participants, or were overlooked in terms of their impact upon the group.
In a second experiment, participants were asked to watch either a video of people discussing job candidates, or a video of market trading being undertaken. Half of the volunteers were directly told about any conflicts among the group discussing the job candidates. At the end of the exercise, the volunteers were asked to evaluate their group or market according to a number of attributes, such as transparency, efficiency, fairness, and integrity.
The results were very similar to the first experiment, with people tending to perceive the groups as far more fair and transparent than the markets.
Lack of trust
To understand why this lack of trust existed, the researchers conducted a third experiment, in which volunteers were recruited from a large forecasting project that often used information markets alongside team and individual predictions.
The volunteers engaged in an experiment similar to the second experiment described above, with a clear halo effect visible, whereby participants trusted the institution they were most familiar with, which in this case was committees. As people developed more exposure to prediction markets, however, so too did their trust in them grow.
“It’s hard at first to trust abstract mechanisms like markets,” the researchers say. “But our research shows that markets are reliable and less susceptible to bias. Large organizations could benefit from using information markets.”
The researchers believe their findings are especially important at a time in which working from home has become the normal, and therefore face to face meetings have become very rare.
“Perhaps as people become more comfortable making business decisions in an environment of decreased interpersonal contact and increased reliance on technology decision markets will be seen as less threatening and find wider use in American organizations,” the authors conclude.
